STATE-OF-THE-ART REVIEW IN ENDOCRINOLOGY
Neuroendocrine regulation of reproduction by GnRH neurons: multidisciplinary studies using a small fish brain model
2023 Volume 70 Issue 4 Pages 343-358
Details
2023 Volume 70 Issue 4 Pages 343-358
After the discovery of GnRH, GnRH neurons have been considered to represent the final common pathway for the neural control of reproduction. There is now compelling data in mammals that two populations of kisspeptin neurons constitute two different systems to control the episodic and surge release of GnRH/LH for the control of different aspects of reproduction, follicular development and ovulation. However, accumulating evidence indicates that kisspeptin neurons in non-mammalian species do not serve as a regulator of reproduction, and the non-mammalian species are believed to show only surge release of GnRH to trigger ovulation. Therefore, the GnRH neurons in non-mammalian species may offer simpler models for the study of their functions in neuroendocrine regulation of reproduction, especially ovulation. Our research group has taken advantage of many unique technical advantages of small fish brain for the study of anatomy and physiology of GnRH neurons, which underlie regular ovulatory cycles during the breeding season. Here, recent advances in multidisciplinary study of GnRH neurons are reviewed, with a focus on studies using small teleost fish models.
Neuroendocrine control of GnRH release in two different modes, episodic and surge release, has been extensively studied in mammalian species such as rodents, sheep, primates, and reflex ovulators. There is a general consensus in mammals that the two modes of GnRH release primarily underlie the so-called negative feedback effects of estradiol leading to follicular development and the positive estradiol feedback leading to the initiation of ovulation [1-3]. Furthermore, the core neuroendocrine components of these feedback system have intensively been studied in recent years. The kisspeptin neurons in the rodent arcuate nucleus that also contain neuropeptides neurokinin B and dynorphin (so-called KNDy neurons) project to “dendrons” of GnRH neurons and constitute the GnRH pulse generator circuitry. On the other hand, those in the rostral periventricular area of the third ventricle (RP3V, also called anteroventral periventricular nucleus, AVPV) of rodents and those in the mediobasal hypothalamus (MBH) of sheep and primates project to cell bodies of GnRH neurons and constitute the GnRH surge generator circuitry [4, 5]. On the other hand, there is a growing body of evidence to suggest that, in non-mammalian vertebrates, pulsatile release of GnRH and LH as observed in mammals does not exist, and the control of follicular development is more heavily dependent on the actions of FSH [6]. In accordance with this view, neurons functionally and anatomically equivalent to the mammalian KNDy neurons have not been identified in fish [7] or other non-mammalian species [8], although kisspeptin, neurokinin B, and other related neuropeptides have been reported to exist in teleosts as well [9, 10]. Furthermore, kisspeptin neuronal system consisting of kisspeptin and its receptor Gpr54 in teleosts have been clearly demonstrated to be dispensable for the control of reproduction, and its evolutionally conserved function may be nonreproductive regulation [11, 12], and the significance of kisspeptin and other related peptides remains less clear in teleosts [7]. On the other hand, there are several lines of evidence to suggest occurrence of surge release of LH in teleosts [13-16], and the surge release of GnRH/LH and its neural mechanisms may be evolutionarily conserved among various vertebrate species. Therefore, in the present review, we confine our discussion to the mechanisms of neuroendocrine regulation of ovulation by GnRH neurons, which have been mainly studied by applying multidisciplinary methods to a small fish brain model.
Ever since the earlier immunohistochemical studies of GnRH neurons, it has not been an easy task to unambiguously demonstrate which GnRH neurons project their axons to the median eminence and release GnRH peptides to the hypophyseal portal vessel [17]. This is mainly because the GnRH neurons in many vertebrate species including mammalian and non-mammalian species are rather sparsely distributed, and thus it has been difficult to trace the axonal trajectories of each GnRH neuron to the median eminence. Furthermore, anatomical studies using retrograde tracer applications into the median eminence or the peripheral circulation revealed that only a portion of the GnRH neurons were labeled, indicating that subsets of GnRH neurons may have another function [17]. Anatomical approaches to identify hypophysiotropic GnRH neurons have been enormously advanced through the development of transgenic (TG) techniques to genetically label GnRH neurons with GFP in rodents [18-20]. However, we should remember that it still remains to be a laborious task to identify the hypophysiotropic nature of GnRH neurons in TG rats and mice, which are rather scattered within the rostral forebrain.
On the other hand, study of the anatomy of hypophysiotropic GnRH neurons in non-mammalian species has undergone different history from that of mammals. It has been generally accepted that up to three paralogous genes for GnRH (gnrh1, gnrh2, and gnrh3) exist in non-mammalian vertebrates, and the three GnRH paralogs have specific expression patterns in the brain [21, 22]. Furthermore, the neurons expressing each GnRH paralog (GnRH1, GnRH2, and GnRH3 neurons) have been shown to have different anatomical and electrophysiological characteristics [23-26]. One of the most interesting such characteristics was reported for the brain of a tropical fish, the dwarf gourami. The GnRH1 neurons in this fish form a rather compact columnar cluster of neurons along the both sides of the ventricle (equivalent to the third ventricle of other species) and their axons form a distinct bundle on each side of the brain to project to the pituitary [26-28], clearly indicating the hypophysiotropic nature of the GnRH1 neurons in the POA. Direct axonal projection of these POA GnRH1 neurons were also demonstrated by retrograde labeling after injection of retrograde tracers to the isolated whole brain-pituitary in vitro preparation, in which the tracer could be almost exclusively taken up by the pituitary cells. This kind of experiment is possible because the hypophysiotropic neurons including the POA GnRH1 neurons in actinopterygian fishes project their axons directly to the pituitary. Thus, retrograde transport of tracers can more easily demonstrate the localization of neurons projecting to the pituitary, and double labeling with GnRH immunohistochemistry or in situ hybridization can label the hypophysiotropic GnRH neurons more reliably, especially when we use the whole brain-pituitary in vitro preparation as described above.
Although not all the teleosts show such distinctive hypophysiotropic axonal projection of POA GnRH1 neurons, there is a general agreement as to the localization of hypophysiotropic GnRH neurons in the POA in teleosts as well as in other non-mammalian species [21]. Such an advantageous anatomical characteristics of hypophysiotropic GnRH neurons have further been strengthened by the development of TG medaka (a small freshwater teleost fish) labeling neurons expressing gnrh1 with GFP [29, 30]. Similar TG fish whose hypophysiotropic GnRH1 neurons are labeled with GFP have also been generated in a cichlid fish as well [31]. TG GnRH-GFP zebrafish has also been generated for the developmental study of GnRH neurons [32]. Unfortunately, however, cyprinid fishes like the zebrafish lack gnrh1 paralog, and another paralog gnrh3 is expressed not only in the hypophysiotropic POA GnRH neurons but also in the neuromodulatory terminal nerve GnRH neurons, which are not hypophysiotropic. Thus, the identification of hypophysiotropic GnRH neurons becomes less easy compared with medaka, and this is one of the reasons why we chose to use medaka, not the zebrafish, for functional analysis of hypophysiotropic GnRH neurons using multidisciplinary techniques.
In medaka, GFP TG animals have already been established for all three paralogs of GnRH (gnrh1, gnrh2, gnrh3) [23, 24, 29, 33, 34], and the distribution of hypophysiotropic GnRH1 neurons have been demonstrated [35] (Fig. 1). The authors examined the localization of the hypophysiotropic GnRH1 neurons using double fluorescence analysis, with retrograde labeling of neurons after biocytin injections to the pituitary [35] (see [36]) and either GnRH immunohistochemistry or in situ hybridization for GnRH1 mRNA. The results clearly demonstrated colocalization of GnRH1 immunoreactivity and retrograde labeling in the ventral population of GnRH1 neurons in the preoptic area (vPOA GnRH1 neurons) (Fig. 1B). On the other hand, they did not find any such colocalization of labelings in the dorsal population of POA GnRH1 neurons (Fig. 1A). Furthermore, the vPOA GnRH1 neurons were shown to have higher expression of gnrh1 mRNA (Fig. 1D). These results in adult medaka were also consistent with the report that GFP-positive axons from the POA project to the pituitary in GnRH1-GFP medaka embryos and larvae [29, 30].
GnRH1 neurons in the ventral POA (vPOA) project directly to the pituitary and express higher level of gnrh1 mRNA than do dorsal POA GnRH1 neurons. A and B, Double fluorescence analysis with retrograde labeling of neurons after biocytin injection to the pituitary (green) and GnRH immunohistochemistry (magenta) demonstrates that the GnRH1 neurons of the vPOA project directly to the pituitary (B). White arrowheads in the merged picture indicate neurons that show colocalization of retrograde tracer and GnRH1 peptide. We did not find any colocalization of retrograde labeling (green) and gnrh1 mRNA (magenta) in the population of dorsal GnRH1 neurons (A). V, Ventricle; IHC, immunohistochemistry. Scale bars, 20 μm. C, Illustrations of a frontal section and a lateral view (left, rostral) of the medaka brain, showing the plane of section corresponding to A, B, and D. Blue squares indicate the dorsal (D) and ventral (V) populations of GnRH1 neurons shown in A, B, and D; note that midline is located in the center of the picture in A. Dm, Area dorsalis telencephali pars medialis; Dl, area dorsalis telencephali pars lateralis; Dp, area dorsalis telencephali pars posterior; POm, nucleus preopticus pars magnocellularis; POp, nucleus preopticus pars parvocellularis; OT, tractus opticus; Vp, area ventralis telencephali pars posterior. D, Representative time-lapse photographs of DIG precipitates visualized for gnrh1 mRNA detection by in situ hybridization in the identical section. Photographs were taken at 10, 30, and 240 min after the application of an alkaline phosphatase substrate, NBT/BCIP. In situ hybridization shows two separate populations of gnrh1 mRNA positive neurons in the area ranging from the telencephalon to the POA. The ventral population of GnRH1 neurons was visualized much earlier during the precipitation, strongly suggesting that the expression of gnrh1 mRNA is higher in the ventral than the dorsal population. Scale bar, 25 μm.
As discussed in the preceding section, the teleost, especially the actinopterygian fishes show a characteristic brain-pituitary relation in which the GnRH neurons directly project their axons to the pituitary to release GnRH, which binds to the gonadotropes (LH and FSH cells) and stimulate release of LH and FSH more or less directly (Fig. 2B, C). This is in contrast to the situations in mammals in which hypophysiotropic portal vessel intervenes between the brain and pituitary, so that in vitro preparation of rodent brain inevitably disconnects its functional connections with the pituitary (Fig. 2A). Further advantage of teleosts is that the two gonadotropins, LH and FSH, are expressed in completely separate cellular populations in the pituitary; in medaka, it has been demonstrated by double labeling in situ hybridization that the fshb and lhb genes (genes for β subunit of FSH and LH, respectively) are expressed in separate pituitary cells [37]. Thus, the control mechanisms of LH can be investigated separately from that of FSH, thereby simplifying the analysis of GnRH control of LH release to induce ovulation. Another experimental advantage of using brains of teleosts, especially small fishes like medaka and zebrafish, is the transparency of the brain and pituitary. This allows us to take out the whole brain out of the skull and make the whole brain in vitro preparation, in which we can perform electrophysiological and Ca2+ imaging experiments while keeping the neural circuitries and brain-pituitary relations almost intact as in in vivo conditions [11, 25, 33-35, 38-41]. Lastly, the experimental advantage of the use of medaka for the neuroendocrine mechanisms controlling ovulation is that the females spawn quite regularly every morning when they are kept under long-day (LD) photoperiod of 14-hour light 10-hour dark (14L10D, breeding condition) with a water temperature around 25–28 ℃ and under good nutrition. Thus, the breeding conditions of medaka can be experimentally controlled by photoperiodicity (LD: breeding condition, and 10L14D short-day (SD) condition: non-breeding), and they are suitable for the study of temporal regulation mechanisms of reproduction [40]. In contrast, the zebrafish spawn many eggs at once and stop spawning for a while after that, so that they may not be advantageous for such a study.
Brain-pituitary relation in mammals (A) and teleosts (B). In mammals, GnRH neurons project their axons to the median eminence and release GnRH peptides to the hypophyseal portal vessel. In teleosts, especially the actinopterygian fishes, show a characteristic brain-pituitary relation in which the GnRH neurons directly project their axons to the pituitary to stimulate release of GnRH. C, In teleosts, axon terminals of GnRH1 neurons release GnRH1 peptide in the pituitary, and GnRH1 peptides bind to GnRH receptors in the pituitary LH cells.
By using whole brain preparation of GnRH1-GFP TG female medaka described above, spontaneous neuronal activities (action currents associated with action potential generation) of the GFP-labeled hypophysiotropic vPOA GnRH1 neurons (Fig. 3A) have been analyzed by using targeted single-unit extracellular loose-patch recordings [35] (Fig. 3B–D). The general pattern of the spontaneous action potential firing was so-called episodic firing and was similar to that of rodent GnRH neurons [42-45].
Electrical activity of the vPOA gnrh1:EGFP neurons in sexually mature female medaka. A, Enlarged view of the cell bodies of vPOA gnrh1:EGFP neurons imaged by conventional fluorescence microscopy. Scale bar, 10 μm. B, Targeted on-cell patch-clamp recording from a vPOA gnrh1:EGFP neuron (for 15 sec). C and D, Spontaneous firing activities of GnRH1-GFP neurons for 1 min in the time period preceding the putative LH surge (at 13:00 h; C) and just before and during the putative LH surge (at 16:00 h; D). Bars above the traces indicate the lighting conditions of the aquarium room; lights-on from 08:00 to 22:00 h (white boxes) and lights-off from 22:00 to 08:00 h (gray boxes). The striped boxes indicate the recording time windows, and the scale bars indicate the action current amplitude and time scale. The table at the bottom of D indicates the recording time groups. E and F, Comparisons of the mean firing rate (spikes/second; E) and the median instantaneous frequency (Hz; F) indicate that vPOA gnrh1:EGFP neurons show higher firing rates and higher median instantaneous frequencies in group e (15:30–19:30 h, n = 12), compared with group d (11:30–15:30 h, n = 11). G, H, Both lhb mRNA and primary transcript show clear time-of-day-dependent changes.
One of the most important findings in the spontaneous electrical activity is that the GnRH1 neurons show time-of-day-dependent changes in the firing frequency (Fig. 3C–F). Since it has been generally accepted that firing frequencies of GnRH neurons are related to GnRH peptide release [42, 46-48], the daily cyclic changes in firing frequency shown here are suggested to underlie the cyclic occurrence of GnRH release, with the GnRH surge in medaka possibly occurring in the evening. However, the authors only analyzed parameters of firing such as mean firing rate (spikes/second), median instantaneous frequency, percent quiescence and maximum duration of quiescence of spontaneous firing, and thus could not show in these experiments that such changes in firing patterns can actually trigger GnRH/LH release, which should be necessary to induce ovulation. Therefore, the relation between the firing frequency and the actual GnRH release was further analyzed by using double TG medaka model in which GnRH1 neurons are labeled with EGFP (these GnRH1 neurons will be called as gnrh1:EGFP neurons) and LH cells are labeled with a genetic Ca2+ indicator protein inverse pericam (IP, gene name IP; the IP-labeled LH cells will be called as lhb:IP cells) [39, 41]. This topic will be further discussed below in the next section, real-time monitoring of GnRH peptide release in the pituitary; high frequency (>6 Hz) firing (HFF) of GnRH neurons triggers GnRH release in the pituitary.
Another important finding in medaka [35] is that levels of expression of lhb mRNA and/or lhb primary transcript in the breeding female also showed time-of-day changes that are comparable to those of firing frequencies of GnRH1 neurons (Fig. 3G, H). Although the serum gonadotropin levels and their daily changes cannot be directly measured because of the small body mass of medaka, the mRNA and primary transcript expression levels provide a good measure of daily changes in the serum gonadotropin levels. These results may suggest that time-of-day changes in the spontaneous firing frequency of GnRH1 neurons cause similar daily changes in the release of GnRH from the pituitary to cause LH surge at the appropriate timing. To summarize, the daily cyclicity in firing frequency of hypophysiotropic GnRH1 neurons was suggested to underlie the cyclic occurrence of GnRH release, with the GnRH surge in medaka possibly occurring in the late evening or at night. Here, the surge release of GnRH1 peptide upregulates lhb mRNA expression over several hours, and LH synthesized in the LH cells are considered to be released during the LH surge on the next day (Fig. 4).
Working hypothesis for the temporal regulation of the ovulatory cycle in the female medaka. The GnRH1 neuronal activity is upregulated to cause HFF as defined in the text in the evening of day 1. HFF of GnRH1 neurons cause release of GnRH from their axon terminals in the pituitary, which immediately stimulates LH cells via GnRH receptors to release LH from the pituitary LH cells. Synchronized release of LH, which occurs via certain mechanisms (see text) brings about the LH surge, which triggers ovulation during the night. At the same time, GnRH1 peptide upregulates lhb mRNA expression over several hours, and LH synthesized and stored in the LH cells will be released during the LH surge on day 2. FSH, follicle-stimulating hormone; PT, primary transcript.
The analysis of the relation between the neural activities of hypophysiotropic GnRH neurons and the release activities of GnRH peptides from the axon terminals is essential for the understanding of the neuroendocrine regulatory mechanisms of ovulation by GnRH neurons. In spite of the importance of this kind of analysis, it has been technically difficult to perform such analysis. Although one report in a teleost fish detected GnRH release in the pituitary using carbon fiber amperometry [49] and another report detected GnRH release in brain slices of adult mice using fast-scan cyclic voltammetry [50], these methods did not allow simultaneous recoding of GnRH neuronal activities and GnRH release at their terminals.
This problem was first tackled by generating TG medaka lines that express a genetic Ca2+ indicator protein IP specifically in the pituitary LH cells to visualize Ca2+ signals in the LH cells to detect GnRH binding to GnRH receptors and consequent increase in [Ca2+]i for the release of LH from the LH cell [41]. Since the brain-pituitary of medaka is rather small and transparent, the authors succeeded in detecting specific Ca2+ signals in response to GnRH peptide applications using almost intact in vitro whole brain-pituitary preparations where the functional relation between the brain and pituitary is maintained. They further generated gnrh1:EGFP;lhb:IP double TG medaka by crossing the two lines of medaka to visualize the gnrh1:EGFP neurons and their axon bundles under the fluorescence microscopy of GFP fluorescence in addition to labeling LH cells with IP (Fig. 5A, B). Then, they recorded Ca2+ fluctuations from the LH cells while electrically stimulating unilateral GnRH1 fiber bundle with a monopolar electrode (Fig. 5B). The electrical stimulation (500-μs pulses at +250 μA every 20 ms for 20 trains) induced a [Ca2+]i rise in LH cells (Fig. 5C), and prior application of GnRH receptor antagonist, analog M, disrupted this response (Fig. 5C, D), indicating release of GnRH peptides from the axon terminals of GnRH neurons in the pituitary. Thus, by taking advantage of the whole brain-pituitary preparation of the double TG medaka, they succeeded in developing a tool to detect GnRH release in real time from the axon terminals of GnRH neurons in the pituitary [41].
Release of intrinsic GnRH by stimulating GnRH neuron axons induces Ca2+ rise in LH cells. (A) Morphological evidence that vPOA GnRH1 neurons directly project to the LH cells in the pituitary. (a) Schematic illustration of medaka whole brain from the ventral side. The blue square in (a) is magnified in (b). (b) Enlarged view of the POA and pituitary of gnrh1:EGFP TG medaka, indicating that GnRH1 fibers (white arrowheads) directly project to the pituitary LH cell area (green fibers in the pituitary). (c) Magnified view of the pituitary in (b). For the axonal stimulation of GnRH1 neurons, we placed a stimulation electrode on the bundle of gnrh1:EGFP axons at the yellow arrowhead. (B) Schematic illustration of the experimental setup for the electrical stimulation and Ca2+ imaging. We placed the electrode (black triangle) right on the entry of the GnRH1 fiber bundle into the pituitary. The pink area in the pituitary indicates the LH cell area. We recorded Ca2+ response from the square area delineated with gray lines on the right. (C) Representative example of Ca2+ responses after electrical axonal stimulation of GnRH1 neurons. The arrows indicate the timing of electrical stimulation. Each graph shows traces from 12 cells (thin gray line) and an averaged trace (thick black line). The moving average was applied to the traces for every three points. (D) The peak amplitude in three times of repetitive GnRH1 fiber stimulations in the LH cells. Five micromolar analog M was used as an antagonist for GnRH receptor. ***, p < 0.001; *, p < 0.05. pit, pituitary.
Next, by taking advantage of this preparation, criteria for the GnRH1 neuronal activities that can induce the release of GnRH1 peptide from their nerve terminals in the pituitary were analyzed (Figs. 6, 7) [39]. The authors used an excitatory neurotransmitter glutamate to trigger action potentials in gnrh1:EGFP neurons during loose-cell patch clamp recording (Fig. 6A) and determined just supramaximal concentration of glutamate (Glu) by examining action potential firing frequencies to be 1 mM (Fig. 6B, C). Here, it should be noted that the mean instantaneous frequency (reciprocal of the interval between firings converted to frequency in Hz) of gnrh1:EGFP neuron firing was about >6 Hz (Fig. 6C). Furthermore, as shown by the statistically significant rise in the peak fluorescent ratio of Fura 2-loaded GnRH1 neurons (Fig. 6D), application of Glu up to 1 mM concentration to the gnrh1:EGFP neurons caused significant [Ca2+]i rise in gnrh1:EGFP neurons (Fig. 6E) as well as GnRH release in their axon terminals in the pituitary, which can be demonstrated by imaging of lhb:IP cells (see Fig. 7). After the analysis described in Fig. 6, Ca2+ imaging of lhb:IP cells were performed in response to local puffer application of 1 mM Glu to gnrh1:EGFP neurons (Fig. 7A). As already determined (Fig. 6), local puffer application of 1 mM Glu to the visually targeted gnrh1:EGFP neurons also caused their high frequency firing (>6 Hz) (Fig. 7A inset) similar to the application of Glu by perfusion (Fig. 6). Under this experimental condition, Ca2+ imaging of lhb:IP cells demonstrated substantial responses in [Ca2+]i in lhb:IP cells (Fig. 7B, 1st Glu with two arrows). Furthermore, the Ca2+ response in lhb:IP cells evoked by puffer application of Glu to the gnrh1:EGFP cell body was abolished during the prior perfusion of Analog M (Fig. 7B blue bar, Glu with two arrows during prior application of 1μM Analog M) and recovered by washout (Fig. 7B, 2nd, 3rd, and 4th Glu with two arrows). These results clearly indicate that Glu puffer application to the cell bodies of gnrh1:EGFP neurons cause their high frequency firing (>6 Hz) and stimulates endogenous GnRH1 peptide release to the pituitary LH cells to induce [Ca2+]i rise in lhb:IP cells. Because the vast majority of excitatory synapses in the brain use glutamate, these results suggest that glutamatergic excitatory synaptic inputs strong enough to induce high-frequency firing (>6 Hz) in the cell bodies of gnrh1:EGFP neurons can trigger release of endogenous GnRH1 peptide from their axon terminals to trigger [Ca2+]i rise in the pituitary lhb:IP cells, which, in turn, will induce LH release. Therefore, these high frequency firing (>6 Hz) of gnrh1:EGFP neurons will be called as HFF (high frequency firing) of GnRH1 neurons in the next section to indicate functional trigger of GnRH1 neuronal activity that causes GnRH1/LH release.
An excitatory transmitter glutamate increases firing frequency and [Ca2+]i of GnRH1 neurons. A. Schematic illustration of experimental setup for recording from GnRH1 neurons. B. Perfusion of glutamate strongly activates firing activity of GnRH1 neurons. C. The mean instantaneous frequency in the following five groups: control (n = 12), 50 μM glutamate (n = 9), 100 μM glutamate (n = 8), 500 μM glutamate (n = 7), and 1 mM glutamate (n = 7). Error bars represent standard error of the mean. D. Perfusion of glutamate increases [Ca2+]i of gnrh1:EGFP neurons. Representative fluorescence ratio (F340/F380) change of Fura 2 by bath application of glutamate in gnrh1:EGFP neurons. E. The peak ratio (F340/F380) in the following six groups: control, 10 μM, 50 μM, 100 μM, 500 μM, and 1 mM glutamate (n = 6).
Puffer application of glutamate to GnRH1 neurons triggers Ca2+ response of lhb:IP cells, indicating LH release from the pituitary. A. Schematic illustration of experimental setup for recording Ca2+ responses of pituitary lhb:IP cells through activation of GnRH receptors in response to puffer application of glutamate to GnRH1 cell bodies. Inset (top) shows a schematic illustration of the relation between released GnRH peptides and the pituitary LH cell with GnRH receptors. Inset (bottom left) shows the puffer pipette for Glu application, and the patch pipette for recording spontaneous action potentials. The bottom right inset indicates a trace showing the increase in firing frequency of the GnRH1 neuron in response to the Glu application (orange arrow). B. A representative trace showing the Ca2+ response of lhb:IP cells by puffer application of 1 mM glutamate to gnrh1:EGFP cell bodies in the presence of 1 mM Analog M.
One may wonder that spontaneous firing of >6 Hz may not be high enough to evoke peptide release, since peptidergic neurons such as vasopressin neurons show a phasic bursting pattern comprising alternating periods of activity (7–15 Hz) and silence, each lasting tens of seconds. The oxytocin neurons may reach intermittent high frequency firing of up to 50 Hz in advance of milk ejection [51]. On the other hand, hypophysiotropic GnRH neurons recorded in brain slices of mice exhibit spontaneous action potential firing that is often arranged into short-term bursts or long-term patterns that arise from various interburst interval ranging from 150–400 ms, and the bursts are short and consist of two to eight spikes per burst [45]. Thus, the spontaneous firing frequency of GnRH neurons in mice brain slices is rather low. Considering the reduced nature of brain slice preparation (maintenance of only local neural circuitry in slice, removal of afferent inputs, etc.), one paper challenged and succeeded in recording from GnRH neurons of anesthetized mice in vivo, but they found firing patterns similar to those in in vitro brain slices [52]. More recently, optogenetic activation of GnRH neurons with channelrhodopsin (ChR2) showed that 10 Hz stimulation for 2 minutes was the minimum requirement to generate a pulse-like increase of LH in circulation [53]. It is interesting to note that both mice GnRH neurons and medaka GnRH neurons are able to release GnRH to trigger LH release at rather low action potential frequency (6–10 Hz). As the authors discussed [53], the optogenetic GnRH neuron stimulation was performed in anesthetized mice, and it should be interesting to check whether the same results are recapitulated in awake animals as well.
Another line of interesting studies includes labeling of hypophysiotropic GnRH neurons with genetic Ca2+ indicators such as GCaMP and analyze Ca2+ changes in GnRH neuron somata and axon terminals [54]. In this study, the authors found that spike and neuropeptide-dependent mechanisms control GnRH neuron nerve terminal Ca2+ rises and increases in axon terminal excitability, which triggers GnRH peptide release to the portal vessels and underlies pulsatile LH release [54]. It should be interesting to apply this kind of Ca2+ imaging to record spontaneous Ca2+ signals of hypophysiotropic GnRH neuron soma and nerve terminals in freely-behaving animals by using fiber photometry (see [55]) to analyze GnRH neuron activities in soma and nerve terminals during pulsatile and surge release of GnRH. Furthermore, it should be interesting to apply similar analysis to the medaka hypophysiotropic GnRH1 neurons during spontaneous ovulation, and this approach is now under way in our laboratory.
As described in the preceding section, a sophisticated real-time monitoring method for GnRH peptide release in the pituitary, while recording GnRH1 neuron firing activities in the same preparation, has been developed for medaka model [39]. As already described in the section Electrophysiological recording of hypophysiotropic GnRH1 neurons, it has already been shown that hypophysiotropic GnRH1 neurons show time-of-day-dependent changes in average firing frequency (mean firing rate, Fig. 3E, and mean instantaneous frequency, Fig. 3F) [35]. Therefore, in the next step of analysis, the occurrence of HFF was used as one of the key criteria for the analysis of spontaneous GnRH1 neuronal activities, and the effects of estrogen were analyzed [40]. As described in Introduction, accumulating evidence has already established that preovulatory GnRH/LH surge is induced by high level of estrogen in many mammals, and the entire neuroendocrine mechanism of GnRH surge generation has been understood as the positive feedback action of estrogen. It has also been generally accepted in mammals that estrogen receptor α (ERα, coded by the Esr1 gene)-expressing kisspeptin neurons located in RP3V/AVPV of rodents and those in the mediobasal hypothalamus (MBH) of sheep and primates directly receive circulating estrogen and project to hypophysiotropic GnRH neurons to stimulate GnRH release, which results in LH surge. However, as also stated in Introduction, kisspeptin does not play essential roles in the control of GnRH neurons or reproductive functions in non-mammalian vertebrates such as teleosts [11, 12] or avian species [8]. Therefore, in non-mammalian vertebrates, the neuroendocrine mechanisms that trigger a preovulatory GnRH/LH surges after folliculogenesis still remain unclear. In addition to the results of time-of-day-dependent changes in firing frequency of hypophysiotropic GnRH1 neurons [35], information on the daily changes in the blood concentration of estradiol (E2) (E2 concentration of whole blood instead of serum was measured because the blood volume is so small in medaka) has been reported [56]. The study found out that breeding female medaka show diurnal changes in blood E2 concentration; blood E2 increase in the evening to dawn under natural conditions.
Therefore, considering the similar time course of these two phenomena [35, 56], it was hypothesized that E2 upregulates the neuronal activity of the GnRH1 neurons to stimulate the GnRH/LH surges. To analyze effects of E2 on the spontaneous GnRH1 neuron firing activities [40], three groups of medaka were prepared: sham-operated, ovariectomized (OVX), and OVX + E (E2 administrated to the OVX females by feeding food pellets containing 100 ng E2 per day). Electrophysiological analysis of spontaneous action potential firing activities in these groups of medaka in the evening revealed that OVX females showed significantly lower mean firing rate than those of Sham, and the decrease was recovered in OVX + E (2d) (E2 treatment for two days), and OVX + E (3d) (E2 treatment for three days) groups (Fig. 8A). Mean firing rate and mean instantaneous frequency in OVX was also lower than Sham, and the decrease was partially recovered in OVX + E (2d) and OVX + E (3d) groups (Fig. 8B, C). The number of HFFs in OVX females was also smaller than those in Sham and moderately recovered in OVX + E (2d) and OVX + E (3d) females (Fig. 8D) compared to Sham. It should be noted that there were no GnRH1 neurons showing more than five HFFs during the 10-min recording period in OVX and OVX + E (2d) females, whereas 22.2% and 10.5% of GnRH1 neurons showed more than five HFFs in Sham and OVX + E (3d) females, respectively (Fig. 8E). The results indicate the importance of E2 for upregulation of GnRH1 neuronal activity in the evening, especially the occurrence of HFF. It is likely that a long-term effect of estrogen is necessary for the upregulation of GnRH1 neuronal activity. In contrast, when the firing activities were recorded in the morning and compared among Sham, OVX, and OVX + E (2d) females (Fig. 8A'), there were no significant differences in each parameter, such as mean firing rate (Fig. 8B'), mean instantaneous frequency (Fig. 8C') and number of HFFs (Fig. 8D'). In addition, there were no GnRH1 neurons showing more than five HFFs during the 10-min recording period in any groups (Fig. 8E').
Comparisons of recordings of spontaneous firing activities of gnrh1:EGFP neurons in the evening and the morning. A–E. Evening recording. Ovarian estradiol upregulates firing activity of the vPOA gnrh1:EGFP neurons in the evening. Recordings of firing activities of gnrh1:EGFP neurons in Sham (n = 18 cells from six fish), ovariectomized (OVX, n = 26 cells from nine fish), 2-day estradiol (E)-treated OVX (OVX + E [2d], n = 14 cells from five fish) and 3-day E-treated OVX (OVX + E [3d], n = 19 cells from six fish) female medaka are shown. (A) Plots showing the time course of instantaneous frequency of a representative neuron for 10-min recording period in each group. Two representative plots are shown for each group, and the upper right inset in each plot shows traces of firing activities from the cell during 20-s period. The blue bars on the time axis indicate the time range corresponding to the trace in the inset. (B) Mean firing rate. Steel–Dwass test, Sham vs. OVX, p = 2.56 × 10–5; Sham vs. OVX + E (2d), p = 0.126; Sham vs. OVX + E (3d), p = 0.125; OVX vs. OVX + E (2d), p = 0.094; OVX vs. OVX + E (3d), p = 0.013; OVX + E (2d) vs. OVX + E (3d), p = 0.978. (C) Instantaneous firing frequency. Steel-Dwass test, Sham vs. OVX, p = 6.93 × 10–6; Sham vs. OVX + E (2d), p = 0.011; Sham vs. OVX + E (3d), p = 0.005; OVX vs. OVX + E (2d), p = 0.565; OVX vs. OVX + E (3d), p = 0.037; OVX + E (2d) vs. OVX + E (3d), p = 0.825. (D) Number of high-frequency firings (HFF), which was defined as a series of more than three consecutive firings with >6 Hz instantaneous frequency. Steel-Dwass test, Sham vs. OVX, p = 0.002; Sham vs. OVX + E (3d), p = 0.444; Sham vs. OVX + E (2d), p = 0.056; OVX vs. OVX + E (3d), p = 0.133; OVX vs. OVX + E (2d), p = 0.969; OVX + E (2d) vs. OVX + E (3d), p = 0.507. (E) Percentage of vPOA gnrh1:EGFP neurons showing more than five HFFs during the 10-min recording period. Different letters indicate a statistical difference (p < 0.05), and b' indicates a tendency to be different from b (p < 0.1) in (B).
A'–E'. Morning recording. The firing activity of the vPOA gnrh1:EGFP neurons was not affected by ovariectomy, nor estradiol administration in the morning. Recordings of firing activities of vPOA gnrh1:EGFP neurons in Sham (n = 10 cells from five fish), ovariectomized (OVX, n = 12 cells from six fish) and estradiol (E)-treated OVX (OVX + E [2d], n = 13 cells from five fish) female medaka are shown. (A) Plots showing the time course of instantaneous frequency of a representative neuron for 10-min recording period in each group. Two representative plots are shown for each group, and the upper right inset in each plot shows traces of firing activities from the cell during 20-s period. The blue bars on the time axis indicate the time range corresponding to the trace in the inset. (B) Mean firing rate. Steel–Dwass test, Sham vs. OVX, p = 0.460; Sham vs. OVX + E (2d), p = 0.679; OVX vs. OVX + E (2d), p = 0.589. (C) Instantaneous firing frequency. Steel–Dwass test, Sham vs. OVX, p = 0.199; Sham vs. VX + E (2d), p = 0.992; OVX vs. OVX + E (2d), p = 0.572. (D) Number of high-frequency firings (HFF), which was defined as a series of more than three consecutive firings with >6 Hz instantaneous frequency. Steel–Dwass test, Sham vs. OVX, p = 0.619; Sham vs. OVX + E (2d), p = 0.906; OVX vs. OVX + E (2d), p = 0.868. (E) Percentage of vPOA gnrh1:EGFP neurons showing more than five HFFs during the 10-min recording period. n.s., not significant.
These results are consistent with the previous studies indicating that GnRH neuronal activity of the E2-treated OVX mice was activated in the evening, but not in the morning [42]. It should be also noted here that when the spontaneous frequency of Sham in the morning (Fig. 8A' Sham) and that in the evening (Fig. 8A Sham) are compared, it is higher in the evening than in the morning (also compare Fig. 8C Sham and Fig. 8C' Sham). This is in good agreement with the previous report [35], corroborating the fact that HFF of GnRH1 neurons, which is capable of triggering GnRH release in the pituitary [39], occurs more frequently in the evening and provides the basis for GnRH surge. The importance of E2 for the occurrence of HFF was also confirmed by the fact that GnRH1 neuronal activity was low in short-day photoperiod-conditioned females, which is non-breeding because of low E2 concentration in blood [40].
It is interesting to note that, in the morning, E2 failed to upregulate the firing activity of GnRH1 neurons. The result suggests the involvement of additional time-of-day signal(s) for triggering GnRH/LH surges at an appropriate timing, in addition to the facilitatory effects of E2. In rodents, it has been suggested that the suprachiasmatic nucleus of the hypothalamus, which functions as a master circadian clock in mammals, controls the circadian rhythm of LH surge generation via GnRH neurons and/or RP3V/AVPV kisspeptin neurons [57]. Although the neuroendocrine mechanisms for time-of-day signals to stimulate GnRH/LH surges have been well established in rodents, those in many other vertebrates including teleosts still remain unclear so far.
Histological evidence in medaka demonstrated the localization of estrogen receptor subtype esr2a but not esr2b (esr2a and esr2b correspond to mammalian ESR2 or ERβ) in almost all gnrh1:EGFP neurons in the POA [40]. Another histological evidence in medaka using esr1:EGFP TG medaka revealed that esr1-expressing neurons in the POA directly project to the GnRH1 neurons [58]. Concerning the three types of Esrs (Esr1, Esr2a, and Esr2b), which are expressed in teleosts, there are gene KO studies of esr in medaka [59] and zebrafish [60]. A detailed analysis of female medaka reproductive phenotypes in esr single KO lines for each subtype reports that esr2a KO females were completely infertile, but their infertility was suggested to be mainly caused by atretic oviduct, and 75% of esr2a KO females showed normal ovulation. On the other hand, both esr1 KO and esr2b KO females were fertile with normal ovarian maturation and gonadotropin mRNA expression [59]. These results indicate that an E2-induced preovulatory LH surge followed by ovulation occurs in every esr KO female in medaka. Another study using esr KO zebrafish [60] demonstrated that all three esr single KO females were fertile, whereas the further analysis of double or triple KO suggested that Esr2a and Esr2b play an essential role in female reproduction. Therefore, it may be suggested that two or three subtypes of Esrs, not the single subtype of Esr, mediate the positive action of estrogen to activate the reproductive regulation in teleosts. Based on the multidisciplinary studies using physiological, anatomical, and molecular genetic techniques in medaka, a working hypothesis for estrogen receptor (Esr)-and time-of-day signal(s)-mediated regulations of GnRH1 neurons in female is proposed (Fig. 9) [40].
Schematic illustration of a working hypothesis for estrogen receptor (Esr)-and time-of-day signal(s)-mediated regulations of GnRH1 neurons in female teleosts. Elevated circulating 17β-estradiol (E2) released from developed follicles upregulates the firing activity of hypophysiotropic GnRH neurons via Esrs. There are several candidate regulatory pathways of GnRH neurons mediated by two or three Esrs. (1) Esr1-expressing neurons in the POA that directly project to GnRH neurons, (2) Esr2a expressed in the GnRH neuron itself [40], (3) Esr2b-expressing neurons (e.g., Kiss1 neurons in the nucleus ventralis tuberis [70]). These three candidate pathways are not mutually exclusive but may function in combinations. For triggering GnRH/luteinizing hormone surges at an appropriate timing, there should be an additional mechanism that conveys time-of-day signal(s). One possible pathway is that time-of-day signal is first conveyed to Esr-expressing neurons, and the Esr-expressing neurons, in turn, activate the GnRH neurons. The neuroendocrine system described here can control the regular estrous cyclicity of female teleosts.
As discussed above, the brain of small teleost fish offers various experimental advantages as a rather simple model for the study of neuroendocrine regulatory mechanisms of ovulation in females. Recent advances in molecular genetic techniques have highlighted zebrafish and medaka as models for various biological fields of study [61, 62]. In addition to general advantages of small fish as biomedical research models [61], medaka has other advantages; the females spawn quite regularly every morning when they are kept under breeding condition (favorable photoperiod, temperature, and nutrition). Furthermore, the brain-pituitary of medaka is rather small and transparent, and is suitable for recording electrical activities of neurons labeled genetically with GFP and for detecting specific Ca2+ signals in GnRH neurons and gonadotrophs in the pituitary using almost intact in vitro whole brain-pituitary preparations, where the functional relation between the brain and pituitary is maintained. These characteristics have greatly contributed to the studies discussed in the present review as well as other studies related to the neuroendocrine regulation of reproduction, neural control of motivation for sexual behaviors by neuromodulatory actions of GnRH, etc. [21, 23-25, 34, 63]. However, there are many unsolved issues in the study of neuroendocrine control of ovulation in non-mammalian vertebrates including teleosts, which deserve future challenges as discussed below.
LH surge may require synchronizing mechanisms among GnRH neurons and/or LH cellsIt has been reported that the amplitude of GnRH and LH release during the surge enormously exceed that during the pulsatile release, and the GnRH increase at the onset of the LH surge can reach values averaging 40-fold greater than baseline in ewes [64]. Accordingly, synchronization of high-frequency firing of multiple GnRH neurons has been postulated to be essential for the sufficient release of GnRH peptide to induce LH release and subsequent ovulation. The dendritic bundling interactions and shared synapses between GnRH neurons, which are suggested to contribute to the synchronization during pulse generation, have been reported in mice [65]. Similar synchronization mechanisms may also contribute to surge generation. In the cichlid fish, hypothalamic (POA) GnRH1 neurons have been reported to be connected to each other by electrical synapses [31]. It has already been demonstrated that terminal nerve-GnRH3 neurons of the dwarf gourami also show synchronized firings in the cluster via electrical coupling [66]. Another line of evidence indicates the occurrence of electrotonic coupling among the anterior pituitary cells probably mediated by gap junctions in a teleost tilapia [67]. In the pituitary gonadotropic cells (LH and FSH cells) of medaka, spontaneous Ca2+ fluctuations in LH and FSH cells, indicative of simultaneous oscillation of spontaneous fluctuations among them, have been reported [41]. Similarly, anatomical as well as functional demonstration of gap junctional connections have been reported in tilapia/zebrafish gonadotropic cells, especially among LH cells [68]. Gap junctional electrical coupling has also been reported among LH cells or FSH cells (homotypic network) and between LH cells and FSH cells (heterotypic network) in medaka, which are suggested to facilitate quick relay of information to each other [69]. It may be also tempting to assume that HFF of even a small number of GnRH1 neurons and the resultant release of GnRH in the pituitary may be boosted by synchronizing mechanisms among the LH cells as well as via the other types of pituitary cells, such as FSH cells, to bring about the LH surge. Thus, future analysis of the neural mechanisms for possible synchronized activities among hypophysiotropic GnRH neurons and/or pituitary LH cells should contribute to the understanding of the mechanisms for sufficient amount of release of GnRH/LH (surge release) to induce ovulation.
Identity of Esr-expressing neurons that are involved in surge generationIn mammals, a rise in estradiol secretion from the maturing follicle(s) is considered to be the core endocrine signal for ovulation, with the site of estrogen positive feedback being the RP3V/AVPV kisspeptin neurons in rodents and neurons in the MBH of sheep and primates. Compelling evidence also suggests that Esr1 in these kisspeptin neurons play a pivotal role [3]. However, as described in Introduction, kisspeptin neuronal system consisting of kisspeptin and its receptor Gpr54 in teleosts has been clearly demonstrated to be dispensable for the control of reproduction, and its evolutionally conserved function may be nonreproductive regulation [9, 10], although Kiss1 neurons in medaka hypothalamus are sexually dimorphic and express Esr1, and are sensitive to E2 in terms of kiss1 gene expression [10, 70] as well as spontaneous firing activities of Kiss1 neurons [71]. Intriguingly, the immunohistochemically labeled Kiss1 neurons in medaka project their axons to POA where hypophysiotropic GnRH1 neurons are located [12]. Furthermore, by using in situ hybridization, localization of neurons expressing Kiss1 receptors in medaka, gpr54-1 and gpr54-2, were shown to be localized in POA neurons surrounding the gnrh1-expressing neurons. However, double fluorescence in situ hybridization showed that gpr54-1 and gpr54-2 expressing neurons were found “adjacent to” the gnrh1-expressing neurons in POA, and not the GnRH1 neurons themselves. Instead, similar double fluorescence in situ hybridization showed that isotocin and vasotocin neurons simultaneously express gpr54-2 but not gpr54-1 [12]. These results provide anatomical evidence for the direct regulation by Kiss1 neurons of isotocin and vasotocin neurons, but not GnRH1 neurons.
These findings, however, may not completely deny the involvement of “Kiss1 neurons” in the estrogenic control of GnRH1 neurons. This is because the “Kiss1 neurons” may express/synthesize other peptides or transmitters that may regulate GnRH1 neurons directly, in spite of the fact that kisspeptin itself may be dispensable for the control of reproduction. Although the Kiss1 neurons in medaka express Esr1 and are sensitive to E2, Esr1-KO medaka and zebrafish have been reported to be fertile [59, 60], indicating that Esr1 may not play important roles in the control, contrary to what has been demonstrated for mammals [3]. However, as discussed above in section GnRH neurons show HFF in the evening, dependent on E2 and time-of-day signals, we should bear in mind that two or three subtypes of Esrs, not the single subtype of Esr, may mediate the positive action of estrogen to activate the reproductive regulation in teleosts. As discussed thus far, the identity of Esr-expressing neurons that are involved in surge generation is a big challenge for future research.
Functional identity and localization of the time-of-day signalsIn the preceding section, the presence of time-of-day signal working in concert with the estrogenic control was proposed [40]. In rodents, it has been postulated that there is a time-of-day signal that originates in the suprachiasmatic nucleus and activates RP3V/AVPV kisspeptin neurons and GnRH soma and dendrites [4]. Unfortunately, neither the master circadian clock like the SCH nor the circadian clocks have been clearly identified in teleosts, in spite of the fact that teleosts also possess clear rhythmicity [72]. This should be an important issue to be explored in future.
The author would like to thank all the members of the Laboratory of Biological Signaling, Department of Biological Sciences, the Graduate School of Science, the University of Tokyo for continuous collaborations and discussion in all the studies cited herein. Special thanks go to Dr. Kana Ikegami and Chie Umatani of the University of Tokyo (present address: Tokyo University of Agriculture and Technology) for critical reading of the manuscript and preparation of Figures 8 and 9. Special thanks also go to Dr. Tomomi Karigo of Johns Hopkins University for preparation of Figures 2 and 4, and Dr. Masaharu Hasebe of Osaka University for critical reading of the manuscript and preparation of Figures 6 and 7. This work was supported by grants from the Japan Society for the Promotion of Science Grants-in-Aid for Scientific Research (JSPS KAKENHI) Grants 26221104 and 21K06262.
The author declares no conflict of interest.
